Conference csc32::consolemanager

>After upgrading to PCM V1.6 to fix a Console extract problem, we have now come
>across another problem which is basically related. As mentioned in call note
>882.0, the serviced console is outputing data without any <CR>'s, thus there
>are no line terminators.
    
    This same things happens when using the LPS$CONSOLE application to
    manage LPS20 printservers via a pseudo-terminal 

>They have setup a event/filter and action routine. When they test the event,
>the event is picked up by the EVENTLIST window. But cannot display context of
>the event. (error = NO CONTEXT). If they  use MB3 on the system's Icon
>to bring up the multi-line window and double click on the event, it caused the
>C3 display to crashes with a Access Violation. (see below for debug and dump)

>This problem is due to the fact that the output is not formatted into records
>less than 1024 bytes. I can see that you will be able to suppress the ACCVIO
>but the way how the eventlist displays the context ( ie start and count). I
>cannot see how you could extract the event data from the console unless you
>flag the console as a stream type output and specify the number of characters
>in the context start/count fields.
    
    Your right. Since there are no record terminators then *every byte*
    becomes part of the same record which precludes using context. A
    decision was made to not delimt the records every X bytes because we
    would be manipulating the data stream which could break event
    detection.

>Is there anyway around this problem ? 
    
    The correct colution may not be possible and that is we need the
    service node to terminate it's records. Does the service node not send 
    terminators because it uses cursor positon sequences? Can we set it up
    so that it thinks it's using a hardcopy device and thus it'll send 
    terminated records?
    
    
 Could this be overcome by modifying the LTA terminal characteristics?
    
    Not really. The wrap/truncate characteristics only effect output to a 
    terminal and not input not to mention that if you did make it wrap you
    would be manipulating the data stream.


>%SYSTEM-F-ACCVIO, access violation, reason mask=01, virtual address = 20202020,
>        Improperly handled condition, image exit forced.

    
    Looks like we overwrote a pointer as the VA is really ascii blanks.
    This should be pretty easy to fix.
    
    Regards,
      Dan

Hi Dan,

	Thanks for your comments about this situation. I have actually had the
opportunity to see the console in person. The screen output is formatted by screen
cursor control (like our SMG ) and messages are display at the bottom and scoll
upwards until they disappear off the screen.  

	I know that the terminal server port can be setup as type HARD and I will
see is this will help out otherwise I will try to get them to modify the PU/LU
terminal characteritics.

	The worst thing about this is that they said that VCS did not have this
problem and they said that the config has not changed except PCM. 

Regards,
Niguel

    >I know that the terminal server port can be setup as type HARD
    >and I will see is this will help out otherwise I will try to get them to modify
    >the PU/LU terminal characteritics.
    
    Changing the terminal server characteristics probably won't help and
    again if it automagically caused the data to wrap it would wrpa it more
    than likely at the wrong place.
    
    >The worst thing about this is that they said that VCS did not
    >have this problem and they said that the config has not changed except PCM.
    
    Actually VCS had different problems with these kind of systems. Your
    customer was lucky enough to not run into them.  The accvio should be
    easy to fix.
    
    Regards,
       Dan

    I also have a favor to ask. We need a process dump of the C3 to
    thoroughly analyze the prolem. The stack dump doesn't give us enough.
    I need you to do the following on the customers system. Do it from a
    DECterm on the PCM engine:
    
    $ install remove console$image:console$c3
    $ SET PROC/DUMP
    $ CONSOLE C3
    
    Now reproduce the problem and make the C3 crash. You will have a file
    called CONSOLE$C3.DMP in your current directory. Make this file
    available via the e-net.
    
    Regards,
       Dan
    

	Hi Dan,

	The process dump is available to you from GIDDAY::DUMPFILES:[BROOKS.Q61631]CONSOLE$C3.DMP;1
	The customer is now running PCM V1.6-ECO1.

	Thanks,
	Niguel